Computed inverse resonance imaging for magnetic susceptibility map reconstruction

Area of Science:

• Medical Imaging
• Computational Physics
• Biophysics

Background:

• Magnetic susceptibility mapping is crucial for understanding various biological processes and pathologies.
• Traditional MRI techniques face challenges in accurately reconstructing susceptibility sources.
• A robust computational model is needed to improve the fidelity of susceptibility mapping.

Purpose of the Study:

• To introduce a novel computed inverse magnetic resonance imaging (CIMRI) model.
• To develop a two-step computational approach for reconstructing magnetic susceptibility from MRI data.
• To validate the CIMRI model's performance using numerical simulations.

Main Methods:

• The CIMRI model reverses the T2*-weighted MRI (T2*MRI) process in two steps: field map calculation from MR-phase images and susceptibility source calculation from the field map.
• The ill-posed 3D deconvolution problem in susceptibility reconstruction is addressed using Tikhonov-regularized matrix inverse, inverse filtering, and total variation (TV) iteration.
• Numerical simulations were conducted to compare reconstructed susceptibility maps against a predefined source.

Main Results:

• Numerical simulations demonstrated that the split Bregman TV iteration solver achieves high-fidelity reconstruction of susceptibility maps from MR-phase images (spatial correlation ≈ 0.99).
• The split Bregman TV iteration solver exhibits noise reduction, edge preservation, and image energy conservation properties.
• Calibration of the TV iteration program by selecting appropriate regularization parameters is essential for brain susceptibility reconstruction applications.

Conclusions:

• The proposed CIMRI model effectively reconstructs magnetic susceptibility sources from T2*MRI data through a two-step computational process.
• The core challenge of the ill-posed 3D deconvolution problem is successfully addressed by the split Bregman TV iteration algorithm.
• The CIMRI model offers a promising approach for accurate magnetic susceptibility mapping in MRI.